Style guideline/convention for defining binary operator in a class (C++) -

this bit of general question, have classes i'd define binary "operations" on, , there several different ways of doing so. example, suppose have vector class i'd implement addition operation on. on 1 hand, overload '+' operator, have read few sources isn't practice (which begs 1 ask why language feature @ all). can see in many cases why methods preferable operator overloading, vector addition agreed upon , using '+' should natural.

on other hand, define add() method in class. define normal method , use x.add(y) perform x + y, doesn't showcase binary operator i'm not sure if should preferred. define static method, instance vector.add(x, y). finally, define add() friend function of class, (mathematically) natural in opinion bit contrary philosophy of oop. i'm hoping bit of insight methods preferable (and why).

the usual approach define reflexive operators (+=, *=, etc.) members modify object applied to, , define non-reflexive operators (+, *, etc.) non-members create copy of 1 of arguments, use corresponding reflexive operator operation, , return new object result.

while java programmers believe add functions thing, because java doesn't have operator overloading. named operations lead really long , unreadable expressions things should simple.